Recent technological advances in the semiconductor industry have permitted dramatic increases in circuit density and complexity, and commensurate decreases in power consumption and package sizes for integrated circuit devices. Single-chip microprocessors now include many millions of transistors operating at speeds of hundreds of millions of instructions per second to be packaged in relatively small, air-cooled semiconductor device packages. A byproduct of these technological advances has been an increased demand for semiconductor-based products, as well as increased demand for these products to be fast, reliable, flexible to manufacture and inexpensive. These and other demands have led to increased pressure to manufacture a large number of semiconductor devices at an efficient pace while increasing the complexity and improving the reliability of the devices.
One important semiconductor device component that affects the control and the performance of the device is the gate electrode. For example, polysilicon has long been used as the gate electrode of Metal Oxide Semiconductor (MOS) devices, such as MOS Field-Effect Transistors (MOSFETs). To increase the carrier concentration in the polysilicon electrode, polysilicon is typically doped very heavily to be either n-type or p-type. As a result of the high doping, the Fermi level (the average electron energy level) of the polysilicon is fixed at either the conduction or the valence band edge, and so the workfunction of polysilicon gate electrode is fixed as well.
The threshold voltage of a MOSFET having a gate electrode over a channel region is dependent upon the difference between the workfunctions of the gate electrode and the channel region. With a fixed workfunction for the gate material, the MOSFET threshold voltage is typically adjusted by choosing the dopant concentration in the silicon below the gate dielectric (e.g., in the channel region). To achieve this, a technique such as ion implantation is used to introduce a specific amount of dopant with desired depth profile in the channel region (this is sometimes referred to as the “threshold-adjustment implant”). However, the threshold-adjustment implant superimposes on top of another doping profile that is used to suppress current leakage (punch-through-suppression implant). The superposition of both doping profiles causes excess dopant to be situated within the channel region, and so degrades the mobility of carriers and reduces transistor speed. In addition, for MOSFET applications where the channel length would be scaled down to sub-70 nanometers (nm), it becomes very difficult to simultaneously optimize both the punch-through-suppression doping and the threshold-adjustment doping profiles. In addition, the discrete distribution of implanted dopants can cause large deviations in threshold voltage from device to device, for sub-70 nm channel lengths, which represents a fundamental limitation to the scaling of MOSFETs. One method of dealing with this problem is to reduce the doping in the channel. In this case, the gate workfunction may be used to adjust the MOSFET threshold voltage.
Polycrystalline silicon-germanium alloys have been suggested as an alternative gate material for workfunction adjustment. For such applications, however, the adjustment is possible only within a limited range, and is only applicable to p-MOSFETs. Furthermore, these alloys suffer from a drawback of manufacturing a gate electrode from a semiconductor (either polysilicon or poly-silicon-germanium); namely these gate electrodes have an electrical carrier concentration that is usually limited to the order of about 1020 cm−3.
The limited carrier concentration in the gate electrode causes a reduction in carrier concentration (depletion) above the gate dielectric when sufficient gate voltage is applied. The depleted portion of the gate electrode behaves like a dielectric instead of a conductor, and therefore increases the overall dielectric thickness of the transistor. Consequently, gate capacitance is reduced, which in turn reduces the transistor's current drive and speed.
To resolve this issue, materials with higher carrier concentrations (e.g., metals) have been considered as alternatives to polysilicon or polycrystalline silicon-germanium, for the gate electrode. Implementations of such metal gate electrodes have been difficult due to the lack of availability of appropriate workfunction values and/or due to chemical instability. Thus, the use of metallic gate electrodes has been hindered.
This limitation and others discussed herein have been a challenge in the semiconductor industry.